Myrmecologische Nachrichten 7 31 - 42 Wien, September 2005

The coexistence of two large-sized thermophilic desert : the question of niche differentiation in Cataglyphis bicolor and Cataglyphis mauritanica (: Formicidae)

Markus KNADEN & Rüdiger WEHNER

Abstract

In the Tunisian highland steppes the desert Cataglyphis bicolor (FABRICIUS, 1793) and C. mauritanica (EMERY, 1906) share common microhabitat preferences. Endowed with equally sized workers they exhibit the same daily activity patterns within equally sized foraging areas, where they scavenge for the same type and size of food items. Neither do they establish food territories nor do their foragers avoid the vicinity of neighbouring nests, be they conspecific or allospecific. Due to this strong overlap of the ecological niches of C. bicolor and C. mauritanica the question arises whether the coexistence of the two species is a stable one, or whether it just reflects a temporary epi- sode within a colonization process recurring time and again in instable environments. Key words: Cataglyphis, competition, niche differentiation Dr. Markus Knaden & Prof. Dr. Rüdiger Wehner (contact author), Department of Zoology, University of Zürich, Winterthurerstrasse 190, CH- 8057 Zürich, Switzerland. E-mail: [email protected]

Introduction Competition is one of the strongest driving forces of evo- size (HÖLLDOBLER 1976, HANSEN 1978, CHEW & DE lutionary processes. As often stated, in a stable environ- VITA 1980, RETANA & CERDÁ 1994, WETTERER 1995), ment two coexisting species should exhibit at least small colony size (FOITZIK & HEINZE 1999), patterns of nest differences in resource exploitation, i.e., occupy different site distribution (LEVINGS & FRANKS 1982, CUSHMAN & niches (GAUSE 1934, HARDIN 1960, MAY 1976a). It is al. 1988), modes of territory defence (HÖLLDOBLER & only in the absence of competition, e.g., if there is a sur- LUMSDEN 1980), and often most conspicuously temporal plus of food or breeding places that the niches of coexist- avoidance by adopting different activity rhythms (BRIESE ing organisms can overlap (MAY 1976b). This "competi- & MACAULEY 1980, KLOTZ 1984, BESTELMEYER 2000). tive exclusion principle" has been criticized as being not Nevertheless, there are some examples that ecologi- testable and based on circular arguments: "If the species cally similar species of ants (same food guild, equal body exist, then they are sufficiently different; if they do not co- size, equal activity rhythms) do in fact coexist. This co- exist, then clearly they are too similar" (KEDDY 1989). existence is explained as the result of changing environ- The focus of the present study on niche separation in ments or of different susceptibilities to parasites and pre- two sympatric species of desert ants, Cataglyphis bicolor dators (FEENER 1981, RETANA & al. 1991). (FABRICIUS, 1793) and Cataglyphis mauritanica (EMERY, The present study focuses on the interspecific and in- 1906), is to demonstrate to what extent the competitive in- traspecific competition in two sympatric formicine desert teractions of the two sympatric species depend on dif- ants, Cataglyphis bicolor and Cataglyphis mauritanica. ferences in space use, foraging time, and food resources. Members of both species are common, equally sized scav- The result will be that on the basis of these parameters, i.e., engers sharing their habitats in the highland and lowland nest site and foraging ecological parameters, alone the co- steppes of central Tunisia (WEHNER & al. 1994). All Cata- existence of the two species cannot be explained. glyphis species belonging to the C. bicolor species group Studies of ant populations have largely contributed to and the C. altisquamis species group (as C. mauritanica our understanding of interspecific niche separation. They does) forage individually for dead (HARKNESS & have provided ample evidence that competition between dif- WEHNER 1977, WEHNER & al. 1983, RETANA & al. 1986, ferent species correlates with the degree of ecological CERDÁ & al. 1989) and do not exhibit any kind of phero- similarity between the species, because two species using mone-mediated recruitment (WEHNER 1987). In many the same limited resources, such as food or nest sites, parts of Tunisia, C. bicolor lives sympatrically with the compete on a higher level than species with different nec- smaller-sized C. albicans ROGER (1859). Usually dif- essities (BERNSTEIN 1979, SEIFERT 1987, HÖLLDOBLER ferently sized foragers prefer differently sized food items & WILSON 1990). The differences between closely related (for seed-harvesting ants see, e.g., DAVIDSON 1977a, b, sympatric species increase when both species are common HANSEN 1978, WHITFORD 1978a, 1978b, CHEW & DE and belong to the same feeding guild (DELAGE 1968, VITA 1980), and SCHMID-HEMPEL (1983) has shown that BERNSTEIN 1974, CHEW & CHEW 1980). Among the fac- this kind of food-size segregation is the basis of the co- tors that differentiate between such closely related spe- existence of C. bicolor and C. albicans. By the same token, cies of ants are body size and its correlation with prey the coexistence of two equally sized and closely related ants of the same food guild, as it occurs in C. bicolor and C. mass (Mettler AM 50, accuracy: 0.1 mg) and the length mauritanica, should imply a high level of competition be- of its maximum extension. tween the two species and thus should have led to niche differentiation. Microhabitat In the present account we address this question of niche The herb Peganum harmala is the most abundant plant differentiation in C. bicolor and C. mauritanica. In parti- species in the research area. It might provide the foraging cular we ask, whether both species differ in morphometrics ants with shady places. Local accumulations of this plant or in the spatial and temporal use of their microhabitat. also might be an indicator for small-scale differences in Knowing that both species forage individually for dead chemical soil structure or soil humidity. We therefore re- arthropods we still test for differences in their diet con- corded the numbers of this plant in circular areas (radius: cerning to sizes of the food items and species composi- 10 m) around the nest entrances and compared the data tion. We further inquire whether competition, both inter- for 10 C. bicolor nests and 10 C. mauritanica nests specific and intraspecific, results in an overdispersed nest- (Mann-Whitney test). Even though the research area is site distribution or in the spatial avoidance of foragers from homogenously and sparsely covered with vegetation (see neighbouring nests. Finally, if food availability were a lim- above), we tested whether nesting sites of C. bicolor and iting resource in these coexisting populations we would ex- C. mauritanica differed in their physical soil structure. pect some kind of inhibitory effect to occur between the for- The hardness of the soil was analysed by a metal cylinder ager forces of adjacent conspecific and allospecific nests. (mass: 2 kg) falling along a vertical rod from 1.5 m height onto a nail provided with a centimetre scale. The depth of the penetrating nail depended on the hardness of Material and Methods the soil. To determine the hardness of the upper soil layer, All experiments were performed 10 km to the southwest measurements started with the tip of the nail positioned of Kasserine (central Tunisia), where C. mauritanica and on the ground. To obtain information about the lower soil C. bicolor occur sympatrically (see WEHNER & al. 1994: layers, the nail was hammered 12 cm into the ground, fig. 10). The research area (N 35° 07', E 08° 44') belongs and then the test was repeated. Upper and lower soil to the Tunisian high altitude steppe, which receives about structures were analysed for 22 C. mauritanica nests and 470 mm of annual precipitation and is characterized by hal- 13 C. bicolor nests. At each nest four measurements were fa grass (WALTER & LIETH 1967, FRANKENBERG 1979). taken at the corners of a square (side length: 30 cm) The sandy and stony area is continuously but sparsely centred about the nest entrance. Mean penetrations at the (< 10 %) covered with vegetation (Stipa tenacissima [L., surface and at 12 cm depth were compared for both spe- 1758], Peganum harmala [LINNAEUS, 1758], several Poa- cies by using the Mann-Whitney test. ceae). Medium annual temperatures amount to only 15 °C Daily activity (WALTER & LIETH 1967), but during our experimental period maximum soil-surface temperature reached 42 °C Foragers leaving the nest were recorded by a photo-sen- (in early July). sor device that counted only the leaving but not the re- turning ants. By simultaneously using 6 of these devices Morphometrics we were able to measure the daily activity of 6 nests Both C. bicolor and C. mauritanica represent large-sized (3 nests of each species) on the very same day. In total, species of the Cataglyphis. Morphometrical data measurements were taken at 12 C. mauritanica and 12 C. (head width measured in dorsal view just in front of an- bicolor nests. terior eye margin; length of the hind-leg tibia) were ob- In addition we labelled 18 foragers of C. bicolor and tained from 500 specimens per species. Two colonies of 15 foragers of C. mauritanica individually, and recorded each species, i.e. four colonies in total, were excavated and the number of foraging runs per ant and day. the individuals preserved in 70 % ethanol. On a random basis, 250 specimens per colony were taken from these Competition samples by the following procedure: the members of each Nest distribution colony were subdivided into five glass vials. After stirr- All nests of the two species were mapped within an area ing the samples, 50 workers were taken from each vial of 250 m × 350 m (56 nests of C. mauritanica and 24 for the measurements. The Mann-Whitney test was used nests of C. bicolor). We used the map to perform nearest- to compare interspecific samples. In addition we compared neighbour analyses (CLARK & EVANS 1954), in order to species specific and nest specific differences by applying test whether interspecific or intraspecific competition has Dunn's Multiple Comparison test. led to overdispersed nest-site distributions. The nearest- neighbour analysis provides a distributional index R, Food preferences which can range from 0 (perfect aggregation) to 2.15 (per- In an area where several nests of both species occurred fect spacing with a hexagonal distribution). A value of 1 in close distances foragers carrying a food item where is indicative of a random distribution. caught (C. bicolor: 102, C. mauritanica: 101). Foragers and food where stored in absolute ethanol and later dried Foraging areas of colonies and foraging ranges of in a desiccator for six days until the mass of ten focus individuals ants remained stable for one day. For each forager we In order to be able to record the spatial layout of the ants' calculated the head width (which in both species is linear foraging activities we painted grids of white lines around correlated with the forager's dry weight [data not shown]) the nest entrances. Each grid consisted of 6 concentric and for its according food item we measured the dry circles (radii: 5, 10, 15, 20, 25 and 30 m) subdivided into

32 8 (n = 5 focus nests) or 16 (n = 8 focus nests) sectors. On 1000, p = 0.17). However, when we tested not only for day zero all foragers of a focus nest were labelled with a interspecific differences but also for intraspecific differ- nest specific colour dot. The following day we scanned ences between the nests, intraspecific differences some- the grid at 1-hour intervals from 8:00 to 17:00 (standard times exceeded the interspecific ones (C. bicolor nest 1 local time) for marked foragers and recorded their posi- vs. C. bicolor nest 2, phead width < 0.001, ptibia length < 0.001; tions within the segments of the grid. Ants digging at the C. bicolor nest 1 vs. C. mauritanica nest 2, phead width > nest entrance were not counted as foragers and therefore 0.05, ptibia length > 0.05, Dunn's Multiple Comparison test). excluded from the analysis. The scanning procedure per- Hence, coexistence does not lead to competition avoid- formed by two observers took 18 - 20 min. for the whole ance by means of different species-specific worker sizes. grid (scanning rate about 1.2 m2/s). Noteworthy the weak but significant difference in the In addition, for each focus nest the mean values of head width between C. bicolor and C. mauritanica ants scanned ants per day were computed in sectors with and collected from inside the nests (Fig. 1, lower inset) dis- without neighbouring nests. The Wilcoxon-Signed-Ranks appeared when only food carrying foragers were taken test was used to test for differences between the two data into consideration (Fig. 2, lower inset). Therefore not only sets. These tests were performed separately for conspecific the distribution of worker sizes overlap widely but the neighbours and for allospecific neighbours. foraging cohorts of both species are even more alike with Activity and foraging success of 9 non-neighbouring a head width in C. bicolor foragers of 2.15 ± 0.34 mm focus nests of C. mauritanica were recorded on two con- and in C. mauritanica foragers of 2.14 ± 0.29 mm. (C. secutive days. On both days the numbers of ants leaving bicolor foragers vs. C. mauritanica foragers, Mann-Whit- the nest, and the numbers of ants returning to the nest with ney test: p = 0.7). The equal sizes of the foragers directly food and without food were recorded within 20 min. lead to the question whether there are any species-spec- periods at the same time of day (always p.m.). On the ific food preferences at all. second day all ants from nests that were less than 40 m apart from the focus nests were prevented from foraging Food preferences by closing the nest entrances with sand mounds. These Cataglyphis ants usually forage for dead arthropods. Due nest closures were checked continuously and, if necessa- to the extensively overlapping worker sizes in C. bicolor ry, re-established until the end of the recording period. and C. mauritanica significant differences in the ants' The control (first-day) data and the (second-day) data of prey sizes were not to be expected. The analysis of about the competition-exclusion experiments (number of ants 100 food items collected in either species simultaneously leaving the focus nest, number of prey items, and forag- and within the same area confirmed this expectation. There ing efficiency, i.e., the number of prey items brought into was a wide and completely overlapping distribution of the the nest divided by the number of returning ants) were sizes of the food items with respect to both the dry weights compared by using the Wilcoxon-Signed-Ranks test. and the lengths of the items. Neither parameter was cor- related with the head widths of the foragers (Fig. 2, top Results graphs). Furthermore, we could not observe any differen- Morphometrics ces in prey qualities (Tab. 1). In conclusion the composi- Cataglyphis bicolor and C. mauritanica can be distin- tion of the food intake into the colony seems to depend guished easily by the shape of their petiole: while the completely on food availability. petiole of C. bicolor (and all other Cataglyphis belonging Microhabitat to the bicolor species group) is nodiform, that of C. mau- ritanica is cuneiform which prevents the ants from lifting Whenever C. bicolor shares its habitat with C. savignyi the gaster as C. bicolor does. Here we asked the question, (DUFOUR, 1862) – another ant of the C. bicolor species whether both species differ also in other morphometrical group – there is a clear-cut nest site specialization with C. parameters that could account for different foraging abili- bicolor nests being more closely located to food provid- ties. We focused on head width, which at least in C. bicolor ing plants (DIETRICH & WEHNER 2003). Here, we asked and C. albicans is correlated with prey size (SCHMID- whether the same is true for sympatric occurrence of C. HEMPEL 1983), and on leg length (length of the hind-leg bicolor and C. mauritanica. Although the actual study site tibia), which influences the ants' speed and ability to stilt is more homogenous and contains no oases or fruit gardens, above the hot substrate (WEHNER 1983, CERDÁ 2001). as the sympatric area of C. bicolor and C. savignyi does, We excavated 2 C. bicolor nests (1054 and 867 ants per we found some local accumulations of the plant Peganum nest) and 2 C. mauritanica nests (1170 and 723 ants per harmala, which could provide shade and honeydew and nest) and measured head width and length of the hind leg which could also indicate small-scale soil differences. tibia for 250 randomly taken ants of each nest. Both C. Nevertheless, we could not detect any differences in the bicolor and C. mauritanica are monomorphic species with- number of plants around the nests (numbers of Peganum out any morphological caste specialization as shown by harmala within a radius of 10 m: C. bicolor: 70 ± 68, C. the linear correlation of both measured parameters in loga- mauritanica: 50 ± 36, n = 10 nests of each species, Mann- rithmic scale (Fig. 1, central graph). They have an exten- Whitney test: p = 0.35). In addition, by testing the hard- sive overlap in size of their workers with a difference in ness of the soil with a special device (see Materials), the head width (Fig. 1, lower inset: C. bicolor: 1.72 ± 0.40 mm, depth of the penetration of a standard-size nail did not dif- C. mauritanica: 1.61 ± 0.50 mm; Mann-Whitney test, n = fer in both species irrespective whether the upper or lower 1000, p < 0.001) and no difference in the length of tibia soil structure were tested (upper soil: C. mauritanica: 7.6 (Fig. 1, left-hand inset: C. bicolor: 3.35 ± 0.79 mm, C. ± 1.1 cm, n = 22, C. bicolor: 7.3 ± 1.9 cm, n = 13, Mann- mauritanica : 3.28 ± 1.02 mm; Mann-Whitney test, n = Whitney test: p = 0.85; lower soil: C. mauritanica: 2.2 ±

33

Fig. 1: Central graph: correlation between head width and the length of the hind-leg tibia (double logarithmic plot). Blue squares: C. mauritanica, blue regression line: y = 1.02x + 0.7, R2 = 0.98; red squares: C. bicolor, red regression line: y = 1.00x + 0.67, R2 = 0.96. ANCOVA revealed differences in slope (F = 11.2, p < 0.001) and intercept (F = 138.4, p < 0.001) when head width was taken as covariate. Data from C. fortis (grey triangles), C. albicans (grey crosses) and C. cursor (grey circles) are given in addition, in order to illustrate the usually high interspecific variability within the genus Cataglyphis. Lower inset: frequency distribution of the tibia lengths of the hind legs in C. bicolor (red bars, n = 500) and C. mauritanica (blue bars, n = 500). Left-hand inset: frequency distribution of the head widths in C. bicolor (red bars, n = 500) and C. mauritanica (blue bars n = 500).

0.7 cm, C. bicolor: 2.2 ± 1.1 cm, Mann-Whitney test: p = color: 5.2 ± 1.7 runs per day, n = 18, C. mauritanica: 4.6 0.95). Hence, coexistence cannot be explained by the use ± 3.7 runs per day, n = 15, Whitney-Mann test, p = 0.27). of different microhabitats. Therefore, the similar numbers of automatically detected departures for both species seem to be based on similarly Daily activity sized forager forces. In this context, it should be pointed In desert ant assemblages subordinate species often for- out, that this similarity in the size of the forager force is age at high temperatures close to their physiological ther- in accord with the total worker forces of the colonies (see mal limits (CERDÁ 2001). Although both C. bicolor and Morphometrics section). It is also in accord with our C. mauritanica are thermophile and belong to a typical recordings at nests of which all foragers had been label- subordinate genus (CERDÁ & al. 1998, CERDÁ 2001), com- led (C. mauritanica: 141 ± 75 foragers per nest, n = 10 petition could drive the activity peak of a possible lower nests; C. bicolor: 100 ± 11 foragers per nest, n = 4 nests). ranked species closer to the daily thermal maximum. Returning to the question of possible temporal shifts of Therefore, we tested whether C. bicolor and C. mauri- foraging activity as a strategy of avoiding competition tanica differ in their daily activity rhythms. At 11 C. between the two coexisting Cataglyphis species, we can bicolor nests we detected 699 ± 817 (110 - 2974) de- refer to Fig. 3 and conclude that the two species exhibited partures (exits), whereas 740 ± 543 (141 - 1755) depar- similar daily activities. There were no significant activity tures were counted at 12 C. mauritanica nests. In addi- peaks during the course of the day. Clearly, temporal tion, the count of foraging runs of individual foragers re- avoidance cannot account for the coexistence of C. bicolor vealed similar activity patterns in both species (C. bi- and C. mauritanica.

34

Fig. 2: Top graphs: Relationship between head width of the foragers and the size of the food items retrieved by them (double logarithmic plots; blue squares: C. mauritanica, red squares: C. bicolor; left: dry weight of food item (C. m.: correlation coefficient r2 = 0.004, p > 0.05; C. b.: r2 = 0.02, p > 0.05), right: length of the maximum linear extension of food item (C. m.: r2 = 0.03, p > 0.05; C. b.: r2 = 0.02, p > 0.05). Lower inset: frequency distribution of the head widths of the foragers captured outside the nest (blue bars: C. mauritanica, n = 101, red bars: C. bicolor, n = 102) and of individuals taken from entire colonies (two colonies in either species; open blue bars: C. mauritanica, open red bars: C. bicolor).

Tab. 1: Food items collected by C. bicolor and C. mauritanica. SPA: small parts of arthropods, UI: unidentified items, PM: plant material. Chi2: p = 0.95 with SPA, UI and PM being excluded from the analysis.

Species Hymenoptera Orthoptera Coleoptera Diptera Hemiptera Araneae SPA UI PM Σ C. bicolor 24 12 6 8 9 4 38 0 1 102 C. mauritanica 27 12 6 4 12 3 29 6 2 101

35

Fig. 3: Daily foraging patterns of C. mauritanica nests (blue squares, n = 12 with 3 nests measured per day) and C. bicolor nests (red squares, n = 12 with 3 nests measured per day) on four consecutive days (6 June 2000 - 9 June 2000). The lines depict the fraction of the entire foraging force in each 30-min interval of the day (averaged over the three conspecific nests per day). The error bars depict the standard deviation. Foragers leaving the nests were automatically counted by a photo-sensor device, while returning ants were not counted. The black line represents the temperature at 1 cm above ground. The black bar in the second graph indicates the duration of a period of continuous rainfall.

36 Tab. 2: Distribution of Cataglyphis mauritanica nests, C. bicolor nests and the nests of both species combined. C. m.: Cataglyphis mauritanica, C. b.: Cataglyphis bicolor, b. s.: both species, N: number of nests, ρ: nest density [nests/m²], Dr: expected mean distance to the nearest neighbour in a large random distribution with density ρ, D: measured mean distance to the nearest neighbour. The parameter R describes the type of distribution (R = 1, random distribution; R = 0, maximum aggregation; R = 2.15, maximum spacing). The asterisks ** indicate that the zero hypothesis R = 1 can be rejected by p < 0.01; n.s. means p > 0.05.

Total research area Core area

C. m. C. b. b. s. C. m. C. b. b. s.

N 55 24 79 50 17 67

ρ 0.0006 0.0006 0.0022 0.0015 0.0005 0.002

Dr [m]19.31 29.48 16.22 12.79 22.01 11.1

D [m] 14.13 19.12 12.08 12.71 11.51 11.34

R 0.73** 0.65** 0.75** 0.98 n.s. 0.52** 1.02 n.s.

nests were distributed in clusters in both the entire and Fig. 4: Distribution of nests within the test area. Red circles: the core area. Hence, if in the coexisting populations of C. bicolor nests; blue circles: C. mauritanica nests. Con- C. bicolor and C. mauritanica intraspecific or interspeci- tinuous line: total test area; dashed line: core area exhibit- fic competition occurred, it had not resulted in a regular ing the highest nest density. spacing of the nests.

In conclusion, we could not detect any niche differen- Territorial behaviour tiation parameters between the two coexisting species Apart from nest-site distributions, competition could re- with respect to either morphometrics, or food and micro- sult in establishing foraging territories (direct competi- habitat preferences, or temporal activity patterns. Never- tion) or in the foragers avoiding the vicinity of neigh- theless, does foraging competition, if it occurs at all, in- bouring nests, because neighbours might have already fluence the spatial habitat use of both species, as it might depleted their surrounding foraging area (scramble com- be reflected in the distribution of nesting sites within the petition). Therefore we investigated the space use pat- sympatric zone and/or in the distribution of the forager terns of the forager forces of individual nests by marking forces around the nests? foragers of 14 nests with nest-specific colours and repeat- edly scanning the foragers' positions within a circular Nearest-neighbour analysis grid (diameter 60 m) painted around the nest entrances. A total of 79 nests (24 C. bicolor nests and 55 C. mauri- The positions of 969 C. mauritanica foragers from 10 tanica nests) occurred within our 8.8-ha large test area nests and of 305 C. bicolor foragers from 4 nests were (Fig. 4). Despite the apparently homogenous vegetation recorded (Fig. 5A). About 60 to 70 per cent of the for- and soil structure characterizing this area, 85 per cent of agers of either species were found closer than 10 m from the nests were located in only 37 per cent (3.5 ha) of the their nest, while there were less than 10 per cent search- area. Hence nearest-neighbour analyses were performed ing in more than 22 m distance from their nests. If one for both the entire area (8.8 ha, 9.0 nests × ha-1) and the defines the foraging range of a colony as a circular area smaller "core area" (3.5 ha) exhibiting the highest nest to which the whole forager force devotes 90 per cent of -1 density (19.1 nests × ha ). its total search time (WEHNER & al. 1983), the radius of These analyses (parameter R, Tab. 2) provide infor- this circle is 22 m in either species (Fig. 5B) and hence mation about the degree to which the observed distribu- much larger than the mean distance between neighbour- tion of nests departs from a random one. The C. maurita- ing nests of C. bicolor and C. mauritanica (12.1 m in the nica nests as well as the nests of both species together entire area and 11.3 in the core area, Tab. 2). This exten- were located in clusters within the whole area, but ran- sive overlap of neighbouring foraging areas immediately domly distributed within the core area, while C. bicolor stimulates the question, whether Cataglyphis foragers

37

Fig. 5: (A) Spatial distribution of the foraging activities of 14 focus nests (filled blue triangles: C. mauritanica; filled red triangles: C. bicolor). The coloured dots depict the positions of foragers of each focus nest during a one-day recording period (for recording procedure see Methods). Open triangles indicate the locations of Cataglyphis nests lying at a radial distance of ≤ 30 m from each focus nest, i.e. within the coloured circle drawn around each focus nest. (B) Cumulative search frequency as a function of the distance from the nest. Blue line: C. mauritanica (n = 10 nests with 969 observed ants); red line: C. bicolor (n = 4 nests with 303 observed ants); error bars depict the standard deviation. The dotted line signatures mark the distance from the nest up to which 90 % of the total amount of foragers were found. The black arrow denotes the mean distance to the nearest neighbouring nest. avoid the vicinity of neighbouring Cataglyphis nests. was no decrease in the foraging density in the vicinity of Therefore, in the vicinities of several focus nests we map- neighbouring conspecific or allospecific Cataglyphis nests ped the positions of all neighbouring nests within the grid (Fig. 6). Single foragers were even observed to forage and tested if these neighbours influenced the search den- less than 1 m away from neighbouring conspecific and sity of the focus colony. Within the grids of the 10 C. mau- allospecific nests without causing any aggressive attacks. ritanica focus nests we mapped a total of 38 C. mauri- In conclusion, competition in the coexisting populations tanica and 4 C. bicolor neighbouring nests. The 4 C. bi- of C. bicolor and C. mauritanica causes neither any regular color focus nests had a total of 39 C. mauritanica and 4 spacing of neighbouring nests nor any territoriality or spat- C. bicolor neighbouring nests. In either species, there ial avoidance in the foraging behaviour of individual ants.

38

Fig. 6: Influence of neighbouring C. mauritanica nests on the spatial distribution of foraging activities of C. mauritanica nests (blue) and C. bicolor nests (red). Open bars: mean numbers (+ SD) of foragers recorded in sectors devoid of neigh- bouring nests; filled bars: mean numbers (+ SD) of foragers in sectors containing neighbouring nests. The numbers within the bars depict the numbers of sectors around each focus nest that did or did not contain neighbouring nests (with a total number of 8 or 16 sectors per nest, depending on the sectioning of the grid). Italic numbers below the bars depict the total number of neighbouring nests of each focus nest. The numbers of foragers did not differ between sectors with or with- out neighbouring nests (Wilcoxon Signed Ranks test: C. mauritanica, p = 0.44; C. bicolor, p = 0.70).

Finally we tested whether food is a limited resource at Discussion all. We excluded nests from foraging and recorded wheth- er this exclusion of competition affected the foraging ac- Ecological similarity of coexisting species and the con- tivity and foraging efficiency of undisturbed C. mauri- cept of niche differentiation tanica nests. The experiment was run with 9 focus nests. As the data presented in this study show, the two sympat- In total 50 neighbouring nests (2 - 10 nests within a radial ric North African species of individually foraging desert distance of 40 m of each focus nest) were prevented from ants, Cataglyphis bicolor and C. mauritanica, do not only foraging. We could not observe any differences in the share the same morphological characteristics that might number of prey items collected (10 ± 7 food items with be influential in their foraging abilities (head width and active neighbours vs. 7 ± 4 with excluded neighbours, leg length, Fig. 1), but also forage for same sized food Wilcoxon Signed Ranks test: p = 0.29) and in the forag- items (Fig. 2) and overlap widely in their microhabitat ing efficiency of the workers between the two situations preferences, their daily activity (Fig. 3), and the spatial use (15 ± 8 % successful foragers with active neighbours vs. of their foraging area (Fig. 5). 12 ± 7 % with excluded neighbours, p = 0.86). Contrary Coexistence of ecologically similar ant species is not as to the hypothesis that closing adjacent nests increases the uncommon as expected due to GAUSE's principle (1934). foraging activity of a focus nest, the number of ants in PONTIN (1961, 1963) challenges this principle on the fact decreased when the neighbours were prevented from basis of his observation that Lasius niger (LINNAEUS, foraging (62 ± 25 nest leaving ants with active neigh- 1758) and L. flavus (FABRICIUS, 1782) persistently coex- bours vs. 39 ± 14 with excluded neighbours, p < 0.05). ist in spite of their overlapping ecological requirements. In conclusion, in coexisting populations of C. bicolor He concludes that regulating factors others than ecologi- and C. mauritanica intraspecific and interspecific compe- cal similarity might be important for stabilizing the co- tition, if it occurred at all, does not lead either to an iso- existence of these two species. However, SEIFERT (1987) form (overdispersed) spacing of the nests or to the estab- could show that both species differ in their habitat use lishment of foraging territories and any other kind of spat- with L. niger foraging mainly above ground and L. flavus ial foraging avoidance. Finally food does not seem to be being specialized on keeping subterranean aphids. RETANA a limiting factor (at least during early summer, when our & al. (1992) mention low population density and differ- experiments were performed). ent susceptibility to parasites as possible effects regulat-

39 ing coexistence. FEENER (1981) proved the stabilizing the area containing a lot of potential nest sites, competi- effect of Apocephalus flies parasitizing on major workers tion for such sites seems to be unlikely. In addition we of dentata MAYR, 1855), but not on workers of never observed nest usurpation as described by CERDÁ & its sympatric competitor Solenopsis texana EMERY, 1895, RETANA (1998) for two coexisting subordinate ant spe- as being indirect competition. Nevertheless, the large a- cies. Next we asked whether food is a limiting resource. mount of studies addressing problems of niche differen- Competition for food – be it interspecific or intraspecific tiation in coexisting species of ants clearly raises the – can influence ant populations on different levels. Over- question as to what the potential factors are that establish dispersed and random nest distribution patterns may re- the coexistence of ecologically similar species (temporal sult from high and low competition, respectively (CLARK avoidance: STEBAEV & REZNIKOVA 1972, BARONI UR- & EVANS 1954, PETAL 1980, RYTI & CASE 1984, 1986). BANI & AKTAÇ 1981, HÖLLDOBLER 1981, 1986, KLOTZ By transplanting Formica obscuripes FOREL, 1886, and 1984; specialization on differently distributed food items: Dolichoderus taschenbergi (MAYR, 1866) colonies into a DAVIDSON 1977a, 1977b, HÖLLDOBLER & al. 1978; wor- previously unoccupied area, a large amount of the colonies ker size and polymorphism: HÖLLDOBLER 1976, DAVID- emigrated when their next neighbour distances were 5 m, SON 1977a, 1977b, 1978, HANSEN 1978, CHEW & DE while colonies with next neighbour distances of 20 m re- VITA 1980, SCHMID-HEMPEL 1983, RETANA & al. 1992, mained in place (BRADLEY 1972). The Cataglyphis nests WETTERER 1995; microhabitat preferences: JOHNSON 2000, within the research area exhibited random distributions DIETRICH & WEHNER 2003). In the present account, with a slight tendency towards clustering (Tab. 2). Obvi- however, we can do nothing but demonstrate the remark- ously, intraspecific and interspecific competition between able ecological similarity of the two coexisting desert the two Cataglyphis species are not strong enough to re- ants C. bicolor and C. mauritanica by having checked as sult in overdispersed nest-site distributions. many parameters of food resource utilization as possible. Beside colony-site distribution, competition can also Most studies documented differences in at least one affect individual foraging patterns. In their review of for- parameter whenever coexisting species were compared. aging strategies in ants, CARROLL & JANZEN (1973) ex- For example, coexisting seed-eating ants of the genera pect "intense competition for scavenged food items." Even Pogonomyrmex and usually differ either in the though the coexisting Forelius pruinosus (ROGER, 1863), size of their workers (and therefore the size of the pre- depilis FOREL, 1901, and M. mimicus ferred seeds) or in details of their foraging behaviour WHEELER, 1906, show wide niche overlap, intense com- (DAVIDSON 1977b, HÖLLDOBLER & al. 1978, RETANA & petition leads to interference behaviour in I. pruinosum, CERDÁ 1994). A comparison of the sympatric Mediter- whose workers prevent the workers of the competing Myr- ranean species Pheidole pallidula (NYLANDER, 1848) mecocystus species from foraging by a chemical repellent and Tetramorium semilaeve ANDRÉ, 1883, revealed wide (HÖLLDOBLER 1982). Much as Forelius, C. bicolor and C. overlaps in diet, foraging times, and habitat requirements mauritanica are scavengers for dead arthropods, and hence (RETANA & al. 1992). In contrast to the current study, the two competing species could have evolved similar types both species belong to distant genera and therefore are not of interference behaviour, but none was observed. Fur- as closely related as C. bicolor and C. mauritanica, which thermore, competition for food could also lead to de- of course belong to different Cataglyphis species groups creased foraging activities in the vicinity of neighbouring (AGOSTI 1990). Pheidole pallidula and Tetramorium semi- nests, since the neighbours might exhaust the food source laeve largely differ in morphometric parameters with the around their nests. If the foragers avoided the vicinity of former possessing a major caste and the latter being mono- neighbouring nests, merely analyzing the extent of for- morphic. As emphasized by RETANA & al. (1992) in eco- aging areas would not allow one to discriminate between logical comparisons of coexisting species one can never scramble competition and direct competition (e.g. territo- be sure that every relevant resource class has been taken ry defence). Nevertheless, our finding that foraging activ- into account. Nevertheless, the main result of the present ity does not decrease in areas close to neighbouring nests study – the extensive similarities of the sympatric thermo- (Fig. 6) speaks against the existence of foraging territor- philic scavengers C. bicolor and C. mauritanica (same ies and points towards rather low levels of scramble com- microhabitats, equal-sized workers foraging for the same petition. It should be mentioned, however, that due to the type of food at the same locations and same time of day) polydomy of at least C. bicolor we cannot exclude that – remains surprising. neighbouring nests belonged to the same colony. Never- theless, when all foragers of 14 focus nests were marked What are the limiting factors? with a nest-specific colour, we never observed a marked HÖLLDOBLER & WILSON (1990) point out that in a pop- ant leaving one of the neighbouring nests. Therefore, even ulation "one factor is usually limiting, and if it were re- if nests belonged to the same colony, their forager forces moved, the population would increase until a second fac- could be regarded as acting independently. The absence tor became limiting, and so on." The number of potential of territoriality in C. bicolor and C. mauritanica is in ac- nest sites has been proved to be a limiting factor in ant cord with the fact that territoriality requires omnipresence populations (HERBERS 1986, FOITZIK & HEINZE 1998). of the workers within the territory, to keep strangers out We could not find any differences in the microhabitats in and defence costs low (HÖLLDOBLER & LUMSDEN 1980, which C. bicolor and C. mauritanica nest. Neither spe- JACKSON 1984). We counted 140 ± 75 foragers at the C. cies relies on potentially limited nesting sites such as emp- mauritanica nests (n = 10 nests) and 100 ± 11 foragers at ty twigs or flat stones (as Temnothorax does), but both the C. bicolor nests (n = 4 nests). These small forager for- dig their nests in the same sandy and stony ground with ces do not allow for the omnipresence of workers within similar plant cover around. Due to the homogeneity of the rather large foraging areas (about 1500 m2 as calcul-

40 ated by the radius of the foraging range; Fig. 5). Finally, References neither foraging activity nor foraging efficiency of a foc- us colony increases, when all neighbouring potential com- AGOSTI, D. 1990: Review and reclassification of Cataglyphis petitors are excluded from foraging. This observation a- (Hymenoptera, Formicidae). – Journal of Natural History 24: 1457-1506. gain supports our notion that among populations of C. bicolor and C. mauritanica food competition is low or BARONI URBANI, C. & AKTAÇ, N. 1981: The competition for food and circadian succession in the ant fauna of a representative even absent – at least when food is not limited. Anatolian semi-steppic environment. – Mitteilungen der Schwei- In conclusion, competition for nest sites or food items zerischen Entomologischen Gesellschaft 54: 33-56. does not seem to play a major, if any role in stabilizing the co- BERNSTEIN, R.A. 1974: Seasonal food abundance and foraging existence of these sympatric species of desert ant scavengers. activity in some desert ants. – The American Naturalist 108: Could the coexistence be explained by different life 490-498. histories? BERNSTEIN, R.A. 1979: Evolution of niche breadth in popula- tions of ants. – The American Naturalist 114: 533-544. When excavating nests we always found only single de- BESTELMEYER, B.T. 2000: The trade-off between thermal toler- alate queens per nest in C. bicolor, but up to 7 dealate ance and behavioural dominance in a subtropical South Amer- queens in nests of C. mauritanica (at other Tunisian study ican ant community. – Journal of Ecology 69: 998-1009. sites up to 30 such queens were found, WEHNER & al. BRADLEY, G.A. 1972: Transplanting Formica obscuripes and 1994). In ants the number of queens often corresponds to Dolichoderus taschenbergi (Hymenoptera: Formicidae) col- different nest founding strategies, with monogynous spe- onies in Jack Pine stands of south-eastern Manitoba. – The cies founding colonies independently after far-reaching Canadian Entomologist 105: 1525-1528. mating flights and polygynous species spreading slowly BRIESE, D.T. & MACAULEY, B.J. 1980: Temporal structure of by budding (HÖLLDOBLER & WILSON 1977, KELLER an ant community in semi-arid Australia. – Australian Jour- 1991, HEINZE & KELLER 2000). RETANA & al. (1992) dis- nal of Ecology 6: 1-19. cuss different life histories as an alternative factor that CARROLL, C.R. & JANZEN, D.H. 1973: Ecology of foraging by might account for the coexistence of similar species. The ants. – Annual Review of Ecology and Systematics 4: 231-257. Tunisian highland and lowland steppes, in which C. bi- CERDÁ, X. 2001: Behavioural and physiological traits to thermal color and C. mauritanica coexist, are erratically used for stress tolerance in two Spanish desert ants. – Etologia 9: 15-27. growing grain. We never found any Cataglyphis nests on CERDÁ, X. & RETANA, J. 1998: Interference interactions and freshly ploughed land. In case of different colony found- nest usurpation between two subordinate ant species. – Oeco- ing strategies one could hypothesize that small scale co- logia 113: 577-583. existing populations of both Cataglyphis species are just CERDÁ, X., RETANA, J., BOSCH, J. & ALSINA, S. 1989: Daily time frames within a colonization process, in which the foraging activity and food collection of the thermophilic ant freshly ploughed land is first occupied by the fast spread- Cataglyphis cursor. – Vie et Milieu 39: 207-212. ing monogynous species, C. bicolor, followed by a slow- CERDÁ, X., RETANA, J. & MANZANEDA, A. 1998: The role of ly budding polygynous one, C. mauritanica. In order to competition by dominants and temperature in the foraging of reveal such long-term processes, it is necessary to survey subordinate species in Mediterranean ant communities. – colonies at the same study site again. Future investiga- Oecologia 117: 404-412. tions will have to show, whether C. bicolor and C. mau- CHEW, A.E. & CHEW, R.M. 1980: Body size as a determinant ritanica differ not only in the number of their queens but of small scale distributions of ants in evergreen woodland, southeastern Arizona. – Insectes Sociaux 27: 189-202. also in their life histories, with C. bicolor and C. mauri- tanica being the independent and dependent colony foun- CHEW, R.M. & DE VITA, J. 1980: Foraging characteristics of a desert ant assemblage: functional morphology and species ders, respectively. separation. – Journal of Arid Environments 3: 75-83. Acknowledgements CLARK, P.J. & EVANS, F.C. 1954: Distance to nearest neighbor as a measure of spatial relationships in populations. – Eco- We wish to thank Pascal Girod, Michael Greef, and Michael logy 35: 445-453. Schmid for their help in acquiring the field data and CUSHMAN, J.H., MARTINSEN, G.D. & MAZEROLL, A.I. 1988: Den- Susanne Reimann for help with the figures. This work sity-dependent and size-dependent spacing of ant nests: evi- was supported by the Swiss National Science Foundation dence for intraspecific competition. – Oecologia 77: 522-525. (grant no. 31-61844-0) and the G. and A. Claraz Founda- DAVIDSON, D.W. 1977a: Species diversity and community or- tion to R.W. ganization in desert seed-eating ants. – Ecology 58: 711-724. Zusammenfassung DAVIDSON, D.W. 1977b: Foraging ecology and community or- ganization in desert seed-eating ants. – Ecology 58: 725-737. In der tunesischen Hochlandsteppe überlappen sich die Ver- DAVIDSON, D.W. 1978: Size variability in the worker caste of a breitungsgebiete der beiden Wüstenameisen Cataglyphis social (Veromessor pergandei MAYR) as a function of bicolor und Cataglyphis mauritanica. Auch in Gebieten, the competitive environment. – The American Naturalist 112: in denen beide Arten sympatrisch vorkommen, besiedeln 523-532. sie die gleichen Mikrohabitate. In Nahrungspräferenzen DELAGE, B. 1968: Recherches sur les fourmis moissoneuses du und tageszeitlicher Aktivitätsrhythmik zeigen sie keiner- Bassin Aquitain: éthologie, physiologie de l'alimentation. – lei Unterschiede. Diese starke Nischenüberlappung in der Annales des Sciences Naturelles Zoologie et Biologie Ani- Ressourcennutzung wirft die Frage auf, ob und inwieweit male 10: 197-165. die Koexistenz beider Arten stabil ist oder einen vorüber- DIETRICH, B. & WEHNER, R. 2003: Sympatry and allopatry in gehenden Kolonisationsprozess in einem instabilen Habi- two desert ant sister species: how do Cataglyphis bicolor and tat darstellt. C. savignyi coexist? – Oecologia 136: 63-72.

41 FEENER, D.H. 1981: Competition between ant species: outcome MAY, R.M. 1976b: Competition and niche theory. In: MAY, R.M. controlled by parasitic flies. – Science 214: 815-817. (Ed.): Theoretical Ecology. – Blackwell; pp. 114-141. FOITZIK, S. & HEINZE, J. 1998: Nest site limitation and colony PETAL, J. 1980: Intraspecific competition as an adaptation to takeover in the ant Leptothorax nylanderi. – Behavioral Eco- food resources in an ant population. – Insectes Sociaux 27: 279. logy 9: 367-375. PONTIN, A.J. 1961: Population stabilization and competition be- FOITZIK, S. & HEINZE, J. 1999: Non-random size differences be- tween the ants Lasius flavus (F.) and Lasius niger (L.). – tween sympatric species of the ant genus Leptothorax (Hym- Journal of Animal Ecology 30: 47-54. enoptera: Formicidae). – Entomologia Generalis 24: 65-74. PONTIN, A.J. 1963: Further considerations of competition and FRANKENBERG, P. 1979: Tunesien. Ein Entwicklungsland im ma- the ecology of the ants Lasius flavus (F.) and L. niger (L.). – ghrebinischen Orient. – Ernst-Klett Verlag, Stuttgart, 172 pp. Journal of Animal Ecology: 565-574. GAUSE, G.F. 1934: The Struggle for Existence. – William & Wil- RETANA, J., BOSCH, J., CERDÁ, X. & ALSINA, A. 1986: Impor- kins Company, Baltimore, 163 pp. tancia del alimento solido y del alimento liquido en el re- HANSEN, S.R. 1978: Resource utilization and coexistence of three gimen trofico de la hormiga Cataglyphis cursor (Hymeno- species of Pogonomyrmex ants in an Upper Sonoran grass- ptera, Formicidae). – Sessions d'Entomologia 4: 139-146. land community. – Oecologia 35: 109-117. RETANA, J. & CERDÁ X., 1994: Worker size polymorphism HARDIN, G. 1960: The competitive exclusion principle. – Sci- conditioning size matching in two sympatric seed-harvesting ence 131: 1292-1297. ants. – Oikos 71: 261-266. HARKNESS, R. & WEHNER, R. 1977: Cataglyphis. – Endeavour 1: RETANA, J., CERDÁ, X. & ESPADALER, X. 1991: corps- 115-121. es in a temperate grassland: a limited supply. – Holarctic Eco- HEINZE, J. & KELLER, L. 2000: Alternative reproductive strat- logy 14: 63-67. egies: a queen perspective in ants. – Trends in Ecology and RETANA, J., CERDÁ, X. & ESPADALER, X. 1992: Coexistence of Evolution 15: 508-512. two sympatric ant species, Pheidole pallidula and Tetra- HERBERS, J.M. 1986: Nest site limitation and facultative poly- morium semilaeve (Hymenoptera: Formicidae). – Entomo- gyny in the ant Leptothorax longispinosus. – Behavioral Eco- logia Generalis 17: 29-40. logy and Sociobiology 19: 115-122. RYTI, R.T. & CASE, T.J. 1984: Spatial arrangement and diet over- HÖLLDOBLER, B. 1976: Recruitment behavior, home range ori- lap between colonies of desert ants. – Oecologia 62: 401-404. entation and territoriality in harvester ants, Pogonomyrmex. – RYTI, R.T. & CASE, T.J. 1986: Overdispersion of ant colonies: a Behavioral Ecology and Sociobiology 1: 3-44. test of hypotheses. – Oecologia 69: 446-453. HÖLLDOBLER, B. 1981: Foraging and spatiotemporal territories SCHMID-HEMPEL, P. 1983: Foraging ecology and colony struc- in the honey ant Myrmecocystus mimicus WHEELER (Hymeno- ture of two sympatric species of desert ants Cataglyphis ptera: Formicidae). – Behavioral Ecology and Sociobiology 9: bicolor and Cataglyphis albicans. – Unpublished PhD Thesis, 301-314. University of Zurich, Zurich, 126 pp. HÖLLDOBLER, B. 1982: Interference strategy of Iridomyrmex pru- SEIFERT, B. 1987: A model to estimate interspecific competitive inosum (Hymenoptera: Formicidae) during foraging. – Oeco- displacement in ants. – Zoologisches Jahrbuch Systematik logia 52: 208-213. 114: 451-469. HÖLLDOBLER, B. 1986: Food robbing in ants, a form of interfer- STEBAEV, I.V. & REZNIKOVA, J.I. 1972: Two interaction types ence competition. – Oecologia 69: 12-15. of ants living in steppe ecosystem in South Siberia, USSR. – HÖLLDOBLER, B. & LUMSDEN, C.J. 1980: Territorial strategies in Ekologia Polska 20: 103-109. ants. – Science 210: 732-739. WALTER, H. & LIETH, H. 1967: Klimadiagramm Weltatlas. – HÖLLDOBLER, B. & WILSON, E.O. 1977: The number of queens: an Gustav-Fischer Verlag, Jena, 472 pp. important trait in ant evolution. – Naturwissenschaften 64: 8-15. WEHNER, R. 1983: Taxonomie, Funktionsmorphologie und Zoo- HÖLLDOBLER, B. & WILSON, E.O. 1990: The Ants. – Belknap geographie der saharischen Wüstenameise Cataglyphis fortis Press of Harvard University Press, Cambridge, MA, 732 pp. (FOREL 1902) stat. nov. (Insecta: Hymenotpera: Formicidae). HÖLLDOBLER, B., STANTON, R.C. & MARKL, H. 1978: Recruitment – Senckenbergiana biologica 64: 89-132. and food-retrieving behavior in Novomessor (Formicidae, Hym- WEHNER, R. 1987: Spatial organization of foraging behavior in enoptera), I: Chemical signals. – Behavioral Ecology and individually searching desert ants, Cataglyphis (Sahara des- Sociobiology 4: 163-181. ert) and Ocymyrmex (Namib desert). In: PASTEELS, J.M. & JACKSON, D.A. 1984: Ant distribution patterns in a Camerooni- DENEUBOURG, J.L. (Eds.): From individual to collective be- an cocoa plantation: investigation of the ant mosaic hypo- havior in social . – Birkhäuser Verlag, Basel, Boston, thesis. – Oecologia 62: 318-324. pp. 15-41. JOHNSON, A.J. 2000: Habitat segregation based on soil texture WEHNER, R., HARKNESS, R.D. & SCHMID-HEMPEL, P. 1983: and body size in the seed-harvester ants Pogonomyrmex rug- Foraging strategies in individual searching ants Cataglyphis osus and P. barbatus. – Ecological Entomology 25: 403-412. bicolor (Hymenoptera: Formicidae). – Gustav Fischer Ver- KEDDY, P. 1989: Competition. – Chapman and Hall, London, lag, New York, 79 pp. 202 pp. WEHNER, R., WEHNER S. & AGOSTI, D. 1994: Patterns of bio- KELLER L. 1991: Queen number, mode of colony founding, and geographic distribution within the bicolor species group of queen reproductive success in ants (Hymenoptera, Formi- the North African desert ant, Cataglyphis FOERSTER 1850 cidae). – Ethology, Ecology and Evolution 3: 307-316. (Insecta: Hymenoptera: Formicidae). – Senckenbergiana bio- KLOTZ, J.H. 1984: Diel differences in foraging in two ant spe- logica 74: 163-191. cies (Hymenoptera: Formicidae). – Journal of the Kansas WETTERER, J.K. 1995: Forager size and ecology of Acro- Entomological Society 57: 111-118. myrmex coronatus and other leaf-cutting ants in Costa Rica. LEVINGS, S.C. & FRANKS, N.R. 1982: Patterns of nest dispersion – Oecologia 104: 409-415. in a tropical ground ant community. – Ecology 63: 338-344. WHITFORD, W.G. 1978a: Structure and seasonal activity of Chi- MAY, R.M. 1976a: Models for two interacting populations. In: huahuan desert ant communities. – Insectes Sociaux 25: 79-88. MAY R.M. (Ed.): Theoretical Ecology. – Blackwell Scientific Pub- WHITFORD, W.G. 1978b: Foraging in seed-harvester ants Pogo- lications, Oxford, London, Edinburgh, Melbourne, pp. 49-70. nomyrmex spp. – Ecology 59: 185-189.

42